Interpreting VCF files: should remove genotypes where REF and ALT alleles are complements?

Question

[darink]

I have been working with several (human) variant files (vcf.) of patients. The data are from high coverage WGS so quality should be high .

Today, just by accident, I noticed a run of 3 consecutive positions on chr1 where the REF and ALT calls in my vcf file were all complementary bases (e.g. the 1000 genomes genotype would be "T/T" whereas the patient's genotype would be "T/A")

I know that the vcf files have already been run through some standard genotyping pipelines that should have accounted phasing but now I'm getting concerned that I still can't trust the results implicitly.

So I'm wondering if there are any best practices for handling instances such as these? Should I simply remove any such complementary "variants" and carry on?